Our reindex documentation does not mention that it is possible to run reindex on secondaries.

As currently implemented, the compact command which is used for similar reasons, places secondaries into RECOVERING mode which has the benefit of not allowing secondary reads to hit that node while the compact is running. However, the reindex command does not currently place the secondary into maintenance mode (SERVER-7824).

As such, the documentation should mention as a caveat that the secondary should preferably be manually placed into maintenance mode during a reindex so that secondary reads are not impacted.